Unlock enterprise-scale web research automation

Today we’re excited to announce the public preview of Deep Research in Azure AI Foundry—an API and software development kit (SDK)-based offering of OpenAI’s advanced agentic research capability, fully integrated with Azure’s enterprise-grade agentic platform.

With Deep Research, developers can build agents that deeply plan, analyze, and synthesize information from across the web—automate complex research tasks, generate transparent, auditable outputs, and seamlessly compose multi-step workflows with other tools and agents in Azure AI Foundry.

AI agents and knowledge work: Meeting the next frontier of research automation

Generative AI and large language models have made research and analysis faster than ever, powering solutions like ChatGPT Deep Research and Researcher in Microsoft 365 Copilot for individuals and teams. These tools are transforming everyday productivity and document workflows for millions of users.

As organizations look to take the next step—integrating deep research directly into their business apps, automating multi-step processes, and governing knowledge at enterprise scale—the need for programmable, composable, and auditable research automation becomes clear.

This is where Azure AI Foundry and Deep Research come in: offering the flexibility to embed, extend, and orchestrate world-class research as a service across your entire enterprise ecosystem—and connect it with your data and your systems.

Deep Research capabilities in Azure AI Foundry Agent Service

Deep Research in Foundry Agent Service is built for developers who want to move beyond the chat window. By offering Deep Research as a composable agent tool via API and SDK, Azure AI Foundry enables customers to:

• Automate web-scale research using a best-in-class research model grounded with Bing Search, with every insight traceable and source-backed.
• Programmatically build agents that can be invoked by apps, workflows, or other agents—turning deep research into a reusable, production-ready service.
• Orchestrate complex workflows: Compose Deep Research agents with Logic Apps, Azure Functions, and other Foundry Agent Service connectors to automate reporting, notifications, and more.
• Ensure enterprise governance: With Azure AI Foundry’s security, compliance, and observability, customers get full control and transparency over how research is run and used.

Unlike packaged chat assistants, Deep Research in Foundry Agent Service can evolve with your needs—ready for automation, extensibility, and integration with future internal data sources as we expand support.

How it works: Architecture and agent flow

Deep Research in Foundry Agent Service is architected for flexibility, transparency, and composability—so you can automate research that’s as robust as your business demands.

At its core, the Deep Research model, o3-deep-research, orchestrates a multi-step research pipeline that’s tightly integrated with Grounding with Bing Search and leverages the latest OpenAI models:

1. Clarifying intent and scoping the task:
   When a user or downstream app submits a research query, the agent uses GPT-series models including GPT-4o and GPT-4.1 to clarify the question, gather additional context if needed, and precisely scope the research task. This ensures the agent’s output is both relevant and actionable, and that every search is optimized for your business scenario.
2. Web grounding with Bing Search:
   Once the task is scoped, the agent securely invokes the Grounding with Bing Search tool to gather a curated set of high-quality, recent web data. This ensures the research model is working from a foundation of authoritative, up-to-date sources—no hallucinations from stale or irrelevant content.
3. Deep Research task execution:
   The o3-deep-research model starts the research task execution. This involves thinking, analyzing, and synthesizing information across all discovered sources. Unlike simple summarization, it reasons step-by-step, pivots as it encounters new insights, and composes a comprehensive answer that’s sensitive to nuance, ambiguity, and emerging patterns in the data.
4. Transparency, safety, and compliance:
   The final output is a structured report that documents not only the answer, but also the model’s reasoning path, source citations, and any clarifications requested during the session. This makes every answer fully auditable—a must-have for regulated industries and high-stakes use cases.
5. Programmatic integration and composition:
   By exposing Deep Research as an API, Azure AI Foundry empowers you to invoke research from anywhere—custom business apps, internal portals, workflow automation tools, or as part of a larger agent ecosystem. For example, you can trigger a research agent as part of a multi-agent chain: one agent performs deep web analysis, another generates a slide deck with Azure Functions, while a third emails the result to decision makers with Azure Logic Apps. This composability is the real game-changer: research is no longer a manual, one-off task, but a building block for digital transformation and continuous intelligence.

This flexible architecture means Deep Research can be seamlessly embedded into a wide range of enterprise workflows and applications. Already, organizations across industries are evaluating how these programmable research agents can streamline high-value scenarios—from market analysis and competitive intelligence, to large-scale analytics and regulatory reporting.

Pricing for Deep Research (model: o3-deep-research) is as follows: 

• Input: $10.00 per 1M tokens.
• Cached Input: $2.50 per 1M tokens.
• Output: $40.00 per 1M tokens.

Search context tokens are charged input token prices for the model being used. You’ll separately incur charges for Grounding with Bing Search and the base GPT model being used for clarifying questions.  

Get started with Deep Research

Deep Research is available now in limited public preview for Azure AI Foundry Agent Service customers. To get started:

• Sign up for the limited public preview to gain early access.
• Visit our documentation to learn more about the feature.
• Visit our learning modules to build your first agent with Azure AI Foundry Agent Service.
• Start building your agents today in Azure AI Foundry.

We can’t wait to see the innovative solutions you’ll build. Stay tuned for customer stories, new features, and future enhancements that will continue to unlock the next generation of enterprise AI agents.

The post Introducing Deep Research in Azure AI Foundry Agent Service appeared first on Microsoft Azure Blog.

As technology continues to reshape every industry, organizations are transforming with cloud, data, and AI—ultimately resulting in smarter operations, better outcomes, and more impactful customer experiences.

From global enterprises to fast-scaling startups, customers are putting Azure to work in powerful ways. To date, more than 26,000 customer projects have leveraged our offerings—Azure Migrate and Modernize and Azure Innovate—to drive faster time to value.

Introducing Azure Accelerate

As customer cloud and AI adoption needs change, we’re evolving our offerings to match your business priorities. Today, we’re excited to introduce Azure Accelerate, a new simplified offering designed to fuel transformation with experts and investments across the cloud and AI journey. Azure Accelerate brings together the full capabilities of Azure Migrate and Modernize, Azure Innovate, and Cloud Accelerate Factory in one place to assist customers from initial planning to full implementation. With Azure Accelerate, customers can:

• Access trusted experts: The deep expertise of Azure’s specialized partner ecosystem ensures your projects launch smoothly and scale efficiently. You can also choose to augment your project by taking advantage of a new benefit within Azure Accelerate—our Cloud Accelerate Factory. Cloud Accelerate Factory provides Microsoft experts at no additional cost to deliver hands-on deployment assistance and get projects up and running on Azure faster.
• Unlock Microsoft investments: Tap into funding designed to maximize value while minimizing risk. Azure Accelerate helps reduce the costs of engagements with Microsoft investment via partner funding and Azure credits. Microsoft also invests in your long-term success by supporting the skilling of your internal teams. Empower them with free resources available on Microsoft Learn, or develop training programs tailored to your needs with a qualified partner. Azure Accelerate supports projects of all sizes, from the migration of a few servers or virtual machines to the largest initiatives, with no cap on investments.
• Succeed with comprehensive coverage: Navigate every stage of your cloud and AI journey with confidence through the robust, end-to-end support of Azure Accelerate. Start your journey with an in-depth assessment using AI-enhanced tools like Azure Migrate to gain critical insights. Design and validate new ideas in sandbox environments and then test solutions through funded pilots or proof-of-value projects before scaling. When you’re ready, start your Azure implementation by having experts build an Azure landing zone. Then, move workloads into Azure at scale following best practices for migrating or building new solutions.

The Cloud Accelerate Factory is a new benefit within Azure Accelerate and is designed to help you jumpstart your Azure projects with zero-cost deployment assistance from Microsoft experts. Through a joint delivery model with Azure partners, these experts can provide hands on deployment of over 30 Azure services using proven strategies developed across thousands of customer engagements. This benefit empowers customers to maximize their investments by offloading predictable technical tasks to our Factory team, enabling internal teams or partners to focus on the more custom and highly detailed elements of a project.

For those organizations who seek guidance and technical best practices, Azure Accelerate is backed by Azure Essentials, which brings together curated, solution-aligned guidance from proven methodologies and tools such as the Microsoft Cloud Adoption Framework, Azure Well-Architected Framework, reference architectures, and more in a single location.

Realizing business value with Azure offerings

Here are just a few examples of how organizations are turning ambition into action:

• Modernizing for agility: Global financial leader UBS is using Azure to modernize its infrastructure, enhancing agility and resilience while laying the foundation for future innovation. This modernization has enabled UBS to respond more quickly to market and regulatory changes, while reducing operational complexity.
• Unifying data for impact: Humanitarian nonprofit Médecins Sans Frontières UK centralized its data platform using Azure SQL, Dynamics 365, and Power BI. This has resulted in streamlined reporting, faster emergency response, and improved donor engagement—all powered by timely, self-service insights.
• Scaling AI for global reach: Bayer Crop Science, in partnership with EY and Microsoft, built a generative AI assistant using Azure OpenAI and Azure AI Search. This natural language tool delivers real-time agronomy insights to farmers worldwide, helping unlock food productivity and accessibility at scale.
• Enhancing insights with AI: OneDigital partnered with Microsoft and Data Science Dojo through Azure Innovate to co-develop custom AI agents using Azure OpenAI and Ejento AI. This solution streamlined research, saving 1,000 person-hours annually, and enabled consultants to deliver faster, more personalized client insights, improving retention through high-impact interactions.

Get started with Azure Accelerate

Azure Accelerate is designed to fuel your cloud and AI transformation. It’s how you move faster, innovate smarter, and lead in a cloud-first, AI-powered world.

We’re excited to partner with you on this journey and can’t wait to see what you’ll build next with Azure. To get started, visit Azure Accelerate to learn more or connect with your Microsoft account team or a specialized Azure partner to plan your next steps.

The post Introducing Azure Accelerate: Fueling transformation with experts and investments across your cloud and AI journey appeared first on Microsoft Azure Blog.

With all our big announcements at Build, you might think we'd kick back and take a break for a few weeks... but fine tuning never stops! We're wrapping up June with a release of direct preference optimization for the 4.1 family of models, fine tuning available in more regions than ever before, and Responses API support for fine-tuned models. 

GPT-4.1, GPT-4.1-mini support Direct Preference Optimization (DPO) 😍

GPT-4.1 and GPT-4.1-mini now support Direct Preference Optimization (DPO). DPO is a finetuning technique that adjusts model weights based on human preferences. You provide a prompt along with two responses: a preferred and non-preferred example; using this data, you can align a fine-tuned model to match your own style, preferences, or safety requirements.  

Unlike Reinforcement Learning from Human Feedback (RLHF), DPO does not require fitting a reward model and uses binary preferences for training. This makes DPO computationally lighter and faster than RLHF while being equally effective at alignment.  

Global Training, now available globally 🌎

Since Build, we've significantly expanded the availability of Global Training (public preview). If you've been waiting for support in your region, we've added another 12 regions! Look for additional features (pause/resume and continuous fine tuning) and models (gpt-4.1-nano) in the coming weeks.    

New available regions: 

• East US 

• East US 2 

• North Central US 

• South Central US 

• Spain Central 

• Sweden Central 

• Switzerland North 

• Switzerland West 

• UK South 

• West Europe 

• West US 

• West US 3 

Responses API now Supports Fine Tuned Models ☎️

Training is great- but inferencing is what matters most when you want to use your models! Responses API is the newest inferencing API. The Responses API is purpose built for agentic workflows: it supports stateful, multi-turn conversations and allows seamless tool calling, automatically stitching everything together in the background. 

With this update, you can make better use of fine-tuned models in multi-agent workflows: after teaching your model what tools to use, and when, RAPI will keep track of conversations so the model can remember context, shows how the model is reasoning through its answers, and let users check progress while a response is being generated. It also supports background processing (so you don’t have to wait) and works with tools like web search and file lookup—making it great for building smarter, more interactive AI experiences. 

I’m refreshed (and sunburned) after a long holiday weekend. Today wasn’t as packed with meetings as usual, so it was also great to crank out a blog post (in addition to this one), empty my inbox, and take some AI training. And get to read a lot, as you’ll see below.

[article] Arriving at ‘Hello World’ in enterprise AI. Top-down selling of AI into enterprises is a tough route. You also need bottoms-up enthusiasm and wins to get real traction.

[blog] 10 Tools For Building MCP Servers. We’re going to overdo it on MCP servers, aren’t we? Seems inevitable. If you want to join the gold rush, here’s a list of frameworks that get you there faster.

[blog] Building your first AI product: A practical guide. Good insight from an engineering leader at Box who helped build their flagship generative AI product.

[blog] Vibe Coding a 5,000km Bike Race Part II: Production ready with Gemini driven development. Fantastic guidance from Esther here on taking a vibe-coded app and working through the key domains that make it production ready.

[article] Why “I’m Sorry” Are Two of the Strongest Words for Leaders. Real sorrys. Not the pretend “I’m sorry you didn’t like what I said” or “I’m sorry for that, but …” stuff.

[article] How has AI impacted engineering leadership in 2025? Good insights, although AI usage data collected in March is already woefully dated. And I wonder if we’re working off a common definition of “developer productivity.” Probably not.

[book] Agentic Design Patterns. My colleague is writing this book out in the open in a series of Google Docs. Anyone can view or offer suggestions. Terrific content!

[blog] A guide to converting ADK agents with MCP to the A2A framework. Don’t add this sort of machinery until you need it. But when you do, it’s good to know how to do it.

[article] Mastercard’s massive structured data stores drive its success with today’s AI applications. Bravo. Seems like the team at Mastercard have put in the hard work to have a great data foundation that now makes AI and ML useful at scale.

[blog] Batch Mode in the Gemini API: Process more for less. It’s async, with higher limits, lower cost, and a good fit for big jobs where results 24 hours later are fine.

[blog] Ready for Rust? Announcing the Official Google Cloud SDK for Rust. Rust has a lot of fans, and now they have easier access to a great cloud platform.

[article] Research: Executives Who Used Gen AI Made Worse Predictions. Check this out to better understand where to guard against thoughtless acceptance of AI answers.

[blog] From Prompt to Code Part 2: Inside Gemini CLI’s Memory and Tools. I like posts that show how to use features. But with open source projects, you can also show how the underlying code actually works. That’s what we have here. And part 3 which explores safety and extensibility of the Gemini CLI.

[blog] From Open Model to Agent: Deploying Qwen3 ADK agent on Vertex AI Agent Engine. Use open models in an open agent framework, and deploy to a cloud runtime. Sounds good to me.

[blog] Hear a podcast discussion about Gemini’s multimodal capabilities. There’s still so much untapped potential when you have LLMs that can understand the world around it. Great discussion between Logan and Ani.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

I have blogged before about how cool Let’s Encrypt is for getting your web things running under https. However I have just got myself a local kubernetes cluster and it is super easy to spin up new web services with SSL certs.

The basic instructions can be found here but let’s look at what was involved.

First of all lets get Cert Manager installed on kubernetes.

kubectl apply -f https://github.com/cert-manager/cert-manager/releases/download/v1.18.1/cert-manager.yaml

Once the cert manager pods are up, you need to create an issuer which communicates with the lets encrypt API. The first code snippet uses the lets encrypt staging environment to avoid any API limits, the second uses production and uses the cloudflare API to authorize SSL requests.

apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
  name: letsencrypt-staging
spec:
  acme:
    server: https://acme-staging-v02.api.letsencrypt.org/directory
    email: your-email@example.com
    privateKeySecretRef:
      name: letsencrypt-staging
    solvers:
    - http01:
        ingress:
          ingressClassName: nginx

apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
  name: letsencrypt-prod
spec:
  acme:
    server: https://acme-v02.api.letsencrypt.org/directory
    email: your-email@example.com
    privateKeySecretRef:
      name: letsencrypt-prod
    solvers:
      - dns01:
          cloudflare:
            apiTokenSecretRef:
              key: api-key
              name: cloudflare-api-token-secret
            email: your-email@example.com

Now that is all configured all I need to do is update my helm chart and any pod I like can have a sub domain of funkysi1701.com with a lets encrypt SSL cert.

This is a section from my helm chart which defines the domain name to use and what issuer to use for the certificate.

ingress:
  enabled: true
  className: "nginx"
  annotations:
    cert-manager.io/cluster-issuer: letsencrypt-prod
  devHost: helloworld-dev.funkysi1701.com
  testHost: helloworld-test.funkysi1701.com
  tls:
    - secretName: helloworld-dev.funkysi1701.com
      hosts:
        - helloworld-dev.funkysi1701.com
    - secretName: helloworld-test.funkysi1701.com
      hosts:
        - helloworld-test.funkysi1701.com

All I need to do now is add similar code like this to every helm chart I publish and my pod will request a SSL certificate. The only manual step I have is to set up a DNS record pointing to the IP address of my cluster for any domain I want to use.

Conclusion

Setting up Let’s Encrypt with Kubernetes and Cert Manager has made it incredibly easy to secure my web services with SSL certificates. With just a few YAML configurations and some simple Helm chart updates, I can automatically provision and renew certificates for any subdomain I need. This approach not only saves time but also ensures my services are always protected with up-to-date encryption. If you’re running Kubernetes, I highly recommend giving Cert Manager and Let’s Encrypt a try for hassle-free SSL management.

ByteDance, the Beijing-based parent company of TikTok, is laying off 65 workers based in Bellevue, Wash., according to a new filing with the Washington state Employment Security Department.

ByteDance, Inc., is laying off 27 workers, while TikTok, Inc., is laying off 38 employees, according to the filing.

The Chinese tech giant landed in the Seattle region in 2021 and has been growing its footprint in Amazon’s backyard as it bolstered its TikTok Shop online shopping business. TikTok has around 1,000 employees in Bellevue — including former Amazon workers — according to a Bloomberg report last month.

But TikTok has recently cut workers from its U.S. e-commerce unit across three rounds of layoffs since April, Bloomberg reported last week, noting that TikTok has replaced some staff near Seattle with managers connected to China.

“As the TikTok Shop business evolves, we regularly review our operations to ensure long-term success,” a spokesperson with TikTok said in a statement to GeekWire. “Following careful consideration, we’ve made the difficult decision to adjust parts of our team to better align with strategic priorities.” 

TikTok Shop is TikTok’s fastest-growing business, according to Bloomberg, though it has fallen short of recent internal sales targets.

TikTok is planning to roll out a new version of its app for users in the U.S. ahead of a planned sale of TikTok’s U.S. operations, The Information reported on Sunday.

ByteDance is one of more than 100 out-of-town tech companies that have engineering centers in the Seattle region, as tracked by GeekWire. ByteDance has more than 440,000 square feet of space in Bellevue and just opened a new office, DJC reported last month.